Modelling of Emergency Vehicle Demand using Poisson Hurdle Regression Model

Abstract

This research presents a Poisson Hurdle Regression Model to explore the factors influencing emergency vehicle demand. The modelling approach built on recent work on spatial modelling field that deals with an excess count of zeros in the dataset. Specifically, this research examines factors contributing to demand of emergency vehicle using geographical attributes at a 1km by 1km grid level. The spatial elements were derived from the dissemination area-level data that provides various characteristics, identifying neighborhood attributes that enabled a comprehensive hypothesis testing during model estimation stage. Moreover, this research examines the temporal characteristics of the demand and compare how different periods of the day affects emergency vehicle demand. The results of this research reveal that social-demographic characteristics, accessibility (such as distance to park, hospital, and business district) and land use measures (such as percentage share of residential and commercial land uses), can influence emergency vehicle demand.